Iziko ledatha eliqhutywa yi-AI lenza umqolo wekamva lethu ledijithali. Ukuze sihlale siphambili, ukukhawulezisa ukusasazwa kweziko ledatha elilungele i-AI kubalulekile, kwaye eli nqaku lihlola amanqanaba amathathu abandakanyekayo.
I-AI ngoku lilitye lesiseko elitsha lophuhliso lwamashishini kwihlabathi liphela. Le teknoloji isetyenziselwa yonke into ukusuka ekwenzeni imisebenzi yesiqhelo ngokuzenzekelayo ukuya ekuveliseni izimvo ezintsha zeemveliso kunye neenkonzo, kwaye impembelelo yayo kulindeleke ukuba ikhawuleze kuphela.
Ngokwengxelo kaMcKinsey ethi "The State of Artificial Intelligence", kunyaka ophelileyo, ama-65% emibutho kwihlabathi liphela aye adibanise i-AI ubuncinane kumsebenzi omnye weshishini (eli nani kulindeleke ukuba lifikelele kuma-50% ngo-2023). Okwangoku, i-IDC iqikelela ukuba ukuveliswa kwedatha yehlabathi kuya kufikelela kwi-175 ZB kulo nyaka, ngokuyintloko kuqhutywa yi-AI, ukufunda koomatshini, kunye nokucutshungulwa kwedatha ngexesha langempela.
Ngokukhula okukhulu kwemarike yeziko ledatha, i-AI iya kuba ngumqhubi ophambili wokukhula. Ngaba iziseko zakho zikulungele oku?
I-AI kwiZiko leDatha: Utshintsho oluphazamisayo
Ii-aplikeshini ze-AI zanamhlanje zihlala zityhala imida yoyilo lweziko ledatha ezikhoyo. Ukusuka ekusingatheni imithwalo yemisebenzi yangaphakathi yeshishini ngokusekelwe kwii-algorithms zokufunda koomatshini ukuya ekuphuculeni ukusebenza kakuhle kwamandla kunye nokhuseleko ngokusebenzisa iimodeli zokuqikelela, i-AI iqhubela phambili amandla okusebenza akrelekrele eziko ledatha ukuya kwiindawo ezintsha.
Olu tshintsho luxhasa amaziko edatha anoxinano olukhulu axhotyiswe ngamaqela eGPU. La maqela anokusingatha imithwalo enzima yemisebenzi ehambelanayo, ehlangabezana neemfuno zamandla ekhompyutha zoqeqesho lwemodeli kunye noqikelelo.
Nangona kunjalo, akukho modeli inye, ebanzi yolu tshintsho. Isantya sokusetyenziswa kwe-AI siyahluka kwiindawo ezahlukeneyo, amashishini, kunye nezibonelelo, nto leyo eyenza ukuqonda okunzulu kwendlela yophuhliso lweziko ledatha ye-AI kubaluleke kakhulu.
Iziseko zoPhuhliso lweZiko leDatha le-AI: Imbono yeHlabathi
Nazi ezinye zeengcinga ezibalulekileyo:
UMntla Melika uthatha ngaphezulu kwe-40% yesabelo semarike yeziko ledatha yehlabathi kwaye kulindeleke ukuba wonyuse amandla ayo ngokuphindwe kabini nge-2.5 kwiminyaka ezayo.
Amazwe afana neIreland, iDenmark, neJamani aba ziindawo zedatha, ngenxa yemigaqo-nkqubo yerhafu efanelekileyo, unxibelelwano oluqinileyo, kunye nokugxila ekuzinzeni.
Ummandla we-Asia-Pacific kulindeleke ukuba ufikelele kwizinga eliphezulu lokukhula (i-CAGR ye-13.3% ukusuka ngo-2025 ukuya ku-2030), ekhokelwa yiTshayina, iJapan, iIndiya, kunye neSingapore.
Amanqanaba Amathathu Okusebenzisa Iziko Ledatha Eliqhutywa Yi-AI
Ukudibanisa i-AI kwimisebenzi yeziko ledatha kudla ngokuqhubeka ngamanqanaba amathathu:
**Ukulungiswa Kwedatha:** Kule nqanaba, i-AI iqokelela idatha kwizixhobo ezahlukeneyo, ezinje ngedathabheyisi, ii-API, iilog, imifanekiso, iividiyo, ii-sensors, kunye neminye imithombo enokuba yeyexesha langempela okanye engeyoyexesha langempela. Le datha emva koko ilebhile/ibhalwe phantsi; iimpazamo ziyasuswa, kwaye iguqulwa ibe yifomathi enokuqondwa yimodeli ye-AI. Esi sisiseko sokuchaneka kwemodeli kunye nokusebenza kwayo.
**Uqeqesho:** Inkqubo ye-AI iqala ukufundisa imodeli ye-AI indlela yokwenza imisebenzi ngesigaba sokulungiselela idatha. Inethiwekhi ye-neural yemodeli ye-AI ifunda idatha, ukwakheka kwayo, iipatheni zayo, kunye nobudlelwane bayo. Oku kwaziwa njengesigaba sokufunda nzulu. Esi sigaba sifuna indawo yeziko ledatha eline-GPU eninzi, enoxinano olukhulu ukuze kucutshungulwe umthwalo wemisebenzi ye-AI ngoxinzelelo oluncinci.
**Ingqiqo/Ukuzimela:** Imodeli ye-AI iqala ukudibana ngokungenamthungo ne-ecosystem yangaphandle kunye nedatha entsha, isenza izigqibo zokugqibela kunye noqikelelo. Kulapho iziseko ze-AI zifuna khona ukufakelwa kweentambo, ii-feed zedatha zexesha langempela, kunye nokuhlanganiswa kwenkqubo enzulu.
Ukoyisa Imingeni yeZiseko zoPhuhliso ukuze kuxhaswe iZiko leDatha eliqhutywa yi-AI
Ukuze kufezekiswe ukuzimela kwe-AI, kufuneka kujongwane nemingeni emininzi ebalulekileyo.
Uxinano lweZibuko kunye neNdawo yeRakhi
Imithwalo yomsebenzi we-AI idla ngokuxhomekeka kwii-GPU clusters ezidityaniswe ngee-links ezinesantya esiphezulu, ezi-low-latency. Oku kubangela uxinano olukhulu lwee-port, okwandisa kakhulu indawo kunye neemfuno zokupholisa. Uyilo lwe-rack yendabuko alukwazi ukuqhubeka. Ngaphandle kweziseko ezizimeleyo, izixhobo ezisetyenziselwa ukukhawulezisa i-AI zinokuba yingxaki.
Ukhetho lweMidiya eDityanisiweyo
Ukukhetha phakathi kobhedu kunye nefayibha akuseyongxoxo yobuchwephesha—yinto ecwangcisiweyo. Iinethiwekhi ze-AI zifuna i-bandwidth ephezulu kunye ne-latency ephantsi kumgama omde. Ifayibha idla ngokuba lukhetho olukhethwayo kwiindawo ezisebenza kakhulu, kodwa kuphela xa icwangciswe kakuhle kwaye ifakiwe. Iimpazamo apha zinokukhokelela ekuncipheni kwesignali kunye nokulahleka kokusebenza, ngakumbi kwiindawo ezinengxolo, ezinokuphazamiseka okukhulu.
Ukuhlanganiswa kwe-IT ne-BAS/BMS
Amaziko edatha e-AI akrelekrele afuna ukuhlanganiswa okungenamthungo, okusebenzisanayo ngexesha langempela kuyo yonke inkqubo yokwakha, nto leyo eyenza ukuhlanganiswa okunzulu kweenkqubo ze-IT kunye neNkqubo yokuZenzekela kweZakhiwo (i-BAS) kunye neNkqubo yoLawulo lweZakhiwo (i-BMS) kubaluleke kakhulu.
Nangona kunjalo, ukuhlanganiswa kwenkqubo okunjalo kudla ngokuthintelwa zizinto ezininzi: iziseko zophuhliso ezindala, iindlela ezahlukeneyo zolawulo kunye nonxibelelwano, kunye neendawo ezingwevu ezingakhange ziqwalaselwe kudala. Ezi ndawo zineenkqubo ezixhasayo eziphambili ezifana ne-UPS, ii-chillers, usasazo lwamandla, kunye nolawulo lwe-HVAC.
Ukuze kusetyenziswe i-AI ukuze kusetyenziswe amandla ngendlela ekrelekrele, ukupholisa, kunye nokhuseleko ngexesha langempela, inkqubo yekhebula esemgangathweni ibalulekile ukuqinisekisa ukunxibelelana okudibeneyo nokuzinzileyo kwazo zonke izinto kwezi ndawo zimdaka. Ngokwahlukileyo koko, iinkqubo zolawulo eziqhekekileyo kunye nokunxibelelana okungalunganga kwenkqubo kunokukhokelela ngokulula ekonakaleni kokusebenza kunye nemingcipheko enzima efana nokungasebenzi kweshishini.
Njengoko ubukrelekrele bokwenziwa buqhubeka bungena kwiimodeli zoshishino, ulindelo lwenkonzo yabasebenzisi, kunye nemisebenzi yedijithali, amaziko edatha kufuneka aqhubeke kwaye ahambelane nophuhliso.
Njengoko kujongwe utshintsho kushishino, ukujongana nemingeni ngendlela ecwangcisiweyo kuye kwaba lukhetho oluyimfuneko ekugcineni ukhuphiswano lwexesha elide. Izigqibo zangoku zocwangciso lweziseko zophuhliso kunye nolwakhiwo ziya kugqiba ngokuthe ngqo ukuba amaziko edatha anokuziqhelanisa nokuphindaphinda ngokukhawuleza kunye nokwandiswa okuguquguqukayo kwezobuchwepheshe be-AI bexesha elizayo. Ukuhlaziya iziseko zophuhliso kwixesha le-AI ngokuyintloko malunga nokwakha ukuguquguquka kwexesha elide kwamaziko edatha.
UBelden HirschmannUluhlu olupheleleyo lwezisombululo zonxibelelwano lubonelela ngepotifoliyo epheleleyo yemveliso eyenzelwe ngokukodwa iimeko zeziko ledatha le-AI ezifuna kakhulu.
Ixesha leposi: Meyi-09-2026